Building on a strong foundation in traditional catalysis and nanomaterials chemistry, the research team has pioneered the integration of machine learning and large language models (LLMs) into catalyst design and mechanistic analysis.

AI-Driven Mechanistic Studies: By employing self-supervised learning on molecular descriptors, the group investigates reaction intermediates, charge transfer pathways, and interfacial microenvironments. This approach provides interpretable structure–performance relationships for MOL/MOF-derived catalysts.

Closed-Loop Materials Screening: Leveraging a self-developed automated synthesis–characterization platform, combined with Bayesian optimization and machine learning algorithms, the team achieves an end-to-end iterative process—from experimental design and data acquisition to model updating—significantly accelerating the discovery of efficient electrocatalytic and photocatalytic materials.

Chemical Large Language Model Assistant: The group is developing “Copilot-Chem”, an intelligent agent that translates natural language commands into experimental control scripts, data analysis workflows, literature knowledge graphs, and machine learning pipelines. This tool empowers scientists to efficiently complete the entire research workflow, from experiment planning to paper writing.

Through the deep integration of chemistry, materials science, computation, and automation, the group is committed to establishing a new paradigm of intelligent catalysis research for energy and environmental applications.